At the risk of oversimplification, empirical analysis of the economics of smoking is guided by three alternative frameworks: rational addiction, imperfectly rational addiction, and irrational addiction.
At least at first glance, addiction seems to be the last place one would find rational behavior. By ‘rational,’ economists mean that agents use all available information to weigh benefits and costs before making a decision. ‘Available’ does not mean that the agent has perfect information because the cost of search is nonzero and not trivial. For decisions with consequences over many periods, the rational agent considers downstream and current effects, which are discounted to present value. These implicit calculations reflect subjective beliefs about probabilities of various outcomes and utilities associated with each outcome with choices based on relative expected utilities.
Smoking is addictive and addiction has the special property that consumption in one period affects future marginal utility and, hence, future consumption of the good. The seminal contribution to the theory of addiction using a rational framework is Becker and Murphy (1988). In this model, the agent is a rational, forward-looking individual with stable preferences. Becker and Murphy give precision to the concept of addictive behavior as consisting of these properties: Tolerance – higher levels of past consumption reduce the marginal utility of consumption in the current period; withdrawal – there is disutility associated with cessation in consumption of the good; reinforcement – higher consumption of the addictive good yields higher marginal utility which leads to higher consumption of the good.
In their model, the rational agent maximizes finite lifetime utility with consumption of the addictive good, a composite nonaddictive good, and the stock of the addictive good (how addicted the agent is). The model implies that consumption of the nonaddictive good is set where marginal utility of consumption of the good equals the discounted marginal utility of wealth – a standard result. For the addictive good, the marginal utility of consumption equals the discounted marginal cost of consuming a unit of the good. Hence the agent anticipates adverse future health effects of consuming the addictive good now as well as future prices of the good.
Imperfectly Rational Addiction
In an imperfectly rational addiction framework, some assumptions underlying the rational addiction model are relaxed: consistency of time preferences, and accuracy of risk perceptions about the harms potentially caused by addictive behaviors or the probability that consuming addictive goods will lead to the person becoming addicted. Under hyperbolic discounting the person’s discount rate, which applies to both gains and losses, increases as the time for implementing the change becomes closer. The discount rate is constant and exponential (the usual assumption about discounting in economics) for decisions occurring in the future, but is higher for decisions occurring in the present. Thus, if the person considers quitting smoking in 10 years, they use exponential discounting. If the issue is quitting smoking today, they use a higher rate. This has the effect of reducing the present value of the benefit of quitting, i.e., reducing the future cost of adverse health effects. To the extent that the expected benefit is lower, there is less incentive to quit. Because the discount rates depend on the timing of the decision, this behavior is said to be ‘time inconsistent.’
Various studies have used this approach and made many modifications to the basic theory, for example, by distinguishing between sophisticated and naıve hyperbolic discounters. The sophisticates know that they will be time inconsistent, so they employ self-control devices, for example, they may not have cigarettes around the house. Also, people may favor smoking bans or higher excise cigarettes as methods of self-control. Others, the naıfs, do not realize that they have time-inconsistent preferences, and so do not consider that at the later time they will discount the benefits of quitting highly, and so not quit early.
Irrational, Cue-Triggered Addiction
In an irrational framework, individuals do not base decisions on objective comparisons of costs and benefits of specific choices, but rather are swayed by emotions. Addictive goods’ consumption may result from visceral urges provoked by external cues, which lead to impulsive consumption that would not have occurred in the absence of these cues. More generally, decision-making often follows pattern matching rather than an explicit weighting of benefits and costs. Behaviors that cues elicit are part of the pattern matching process.
Characteristic ‘irrational’ assumptions are that consumption among addicts is considered by the addicts themselves as a mistake; that environmental cues based on past experiences trigger consumption; and that addicts understand and manage their susceptibility, i.e., they are ‘sophisticates’. In one model (Bernheim and Rangel, 2004), a person can be in either a ‘cold’ or a ‘hot’ state. When cold, the person is rational. However, when exposed to certain environmental stimuli, for example, places where the individual has often smoked in the past, the person switches to the hot state in which emotions supersede rational decision-making. The person can control the probability of receiving an environmental stimulus by altering their lifestyle activity. However, the utility from a lifestyle that provides a high probability of a stimulus (e.g. attending parties) is higher than the utility from another activity (e.g. in the extreme, rehabilitation).